Abstract
Two retarded boys exhibited abnormally low rates of smiling. In Exp. I, the frequency of a boy's smiling was first increased with candy reinforcement, but the frequency of the response did not decrease when candy reinforcement was terminated. When the subject wore a sign designed to make social interactions contingent on not-smiling, the frequency of smiling decreased. The sign was then changed to make social interactions contingent on smiling and the rate of smiling increased. In Exp. II, a second boy initially never smiled. Establishment of a contingency for candy reinforcement did not increase this zero response rate. Instructing the child to smile initially increased smiling, but the instructions then became progressively more ineffective. Candy reinforcement increased the rate of smiling to a normal range, but the rate of the response promptly decreased when this reinforcement was discontinued. Continuous candy reinforcement was again employed to increase the response rate and then progressively leaner schedules of variable-ratio candy reinforcement were employed. Consequently, the rate of smiling did not decrease when candy reinforcement was again eliminated. Subsequently, signs were employed to regulate social interactions and the rate of smiling was shown to be controlled by these interactions serving as reinforcers.